Should operational research be used to design a cruise liner's buffet?
Oh dear, I wonder if I can ever stop being someone who "wonders about it all"? (And that quotation is not original; one of my undergraduate tutors was the astronomer R.A. Lyttelton, and as students we were amused and impressed in equal measure that this was his hobby in "Who's Who" in the days of printed information like that.)
We have just returned from holiday, our first cruise, and it has been memorable for several reasons. As it was a "first " for us, there were many things to consider and experience. Not all are relevant to this article.
But here is one relevant bit. We ate meals in the principal dining room for passengers - a much smaller one was used for superior meals. Breakfast and lunch were buffet service and, in the evening, we enjoyed being at an assigned table with two other couples. The inevitable introductions followed and so the others teased out of me that wondering about queues and their management was a bit of a habit. So, I was challenged to think about how to improve the buffet service.
Seven years ago, I blogged about stock control in a hotel buffet breakfast room. It is here. Since then I have observed hotel buffets in several countries, in small and large hotels. Tina and I have noticed both good stock control and appalling. And we have observed rather too many badly laid out buffet systems for comfort! However, seeing the problem and working out how to do something are different. There have been quite a few breakfast buffets where fresh loaves of bread were on offer, with one board and one bread knife, at the centre of the buffet display, so this was a bottleneck for those collecting food as they progressed along the display, and a nuisance for anyone wanting bread and having to cross the lines; how much better to have several bread boards and knives in a place which did not get in the way of other people.
On the cruise, the problem was, as usual, congestion. There were two sides to each of two aisles, one side with hot food or salads, the other with sweets, cutlery and soup. The two aisles were not identical. They flanked an aisle where kitchen staff delivered food to the hot trays. You could collect a hot meal from either aisle, but one end of one aisle had cheese, cold meat and cold fish, while the corresponding end of the other aisle had vegetables for salad. Each aisle was just wide enough for three people to squeeze through. And they needed to, as there was no "right way" along any of the four sides. The flow was, putting it mildly, chaotic. But the aisles were fixed, their equipment was fixed, and the only way that the chaos could be eliminated was to rearrange the order of the serving dishes to encourage orderly flow.
In terms of queue theory, each tray of food was a server, and customers were passing through a set of sequential queues, for each of which they could be served or choose to pass. So, the trays of food (not just trays, but other sources of food such as bowls and jars) needed to be arranged in an order which made sense for both the kitchen and the diner. As a very simple example, to make a cup of tea, you had to:
1) Collect a cup from point A
2) Collect a tea bag in a sealed sachet from point B
3) Unwrap the tea bag and dispose of the sachet in a bin at point B
4) Fill the cup with boiling water from a dispenser close to point A and between points A and B
5) Add milk from a jug and sugar (if desired) from near point B
Making coffee was easier, as it involved only step 1, and steps 4(the dispenser had a button for coffee) and 5. It is left to the reader to solve this particular problem.
After being challenged to think about the problem in the dining room, my thoughts turned to the interaction of system design and system management. Long ago, somebody had commissioned the cruise ship, and had specified dimensions, access from the kitchens, power connections and equipment, storage space. But had the system designers considered the day to day system management? They could have iterated through possible designs with simulated consumers - even though the behaviour of those consumers is varied, digital simulation could cope. In other environments, the interaction of design and management is evident. (Famously, London Heathrow terminal 5 is "a building designed around the flow of luggage ".)
So, my solution to the queues in the buffet is that the problems should have been resolved at design stage. But something could be done, by modelling different meals and food choices in the present, fixed, layout of walls, aisles and hot plates, with changed layout of the pinch points.
Using Google scholar, I looked for any published academic studies of the problem, and there seem very few. Maybe the cruise ships and related industries have not employed operational research, or maybe the results are a guarded secret? I wonder about it all!